It claims the ripple tolerances to be sub 20mV...
Thanks, I had seen that once.
But it looks to me more like it comes from the marketing department.
What does 20mV mean? mVpp?
I assume mVrms, which is not exactly low noise.
...and the posted video above tries to highlight getting way below that..
I don't know what he's doing there.
Too fast, too blurred, too vertical, too Chinese...

In addition, I now believe that when looking at or measuring spikes, it only makes sense to make comparisons with the same scope, method and setup.
Even with the same scope, you can get drastically different results depending on the acquisition method, memory depth, time scale, bandwidth limitation, etc.
So it only makes sense if you leave everything the same and then look at different PSUs.